Real-Time Visualization of Large Time-Varying Molecules

摘要

Interactive display of molecular datasets is an important tool to better understand the structural and functional properties of biological samples. It remains a challenge to display large molecular datasets, especially time-varying ones, interactively. In this paper, we develop a time- and memory-efficient algorithm to solve this problem. Our approach speeds up the graphics rendering pipeline at several stages by developing and extending various rendering techniques for efficient display of time-varying molecular data, such as run-time triangle strip and triangle fan generation, visibility-based culling, view-dependent precision control, and memory-bandwidth reduction. More importantly, our algorithm requires no pre-processing and little memory overhead. It is linearly scalable in the sizes of the molecular datasets. Overall, we have achieved interactive display of time-varying molecular datasets with several thousands of atoms. Our algorithm is flexible and scalable and our ideas for this problem can also be applied to visualization of other large time-varying datasets, We illustrate the capability of the proposed algorithm by presenting a fast animation of the gating transition of the bacterial mechanosensitive channel MscL.